Benchmarking van der Waals-treated DFT: The case of hexagonal boron nitride and graphene on Ir(111)

TL;DR

This paper evaluates various van der Waals-inclusive density functional theory methods by benchmarking their ability to predict the atomic geometries of hexagonal boron nitride and graphene on Ir(111), revealing significant variability and limited predictive accuracy.

Contribution

It provides a comprehensive benchmark of vdW-corrected DFT methods against experimental data for 2D material interfaces, highlighting current limitations in predictive reliability.

Findings

Significant variation in predicted geometries across methods

Some methods match experimental structures post hoc

Current vdW-DFT approaches lack consistent predictive power

Abstract

There is enormous recent interest in weak, van der Waals-type (vdW) interactions due to their fundamental relevance for two-dimensional materials and the so-called vdW heterostructures. Tackling this problem using computer simulation is very challenging due to the non-trivial, non-local nature of these interactions. We benchmark different treatments of London dispersion forces within the density functional theory (DFT) framework on hexagonal boron nitride or graphene monolayers on Ir(111) by comparing the calculated geometries to a comprehensive set of experimental data. The geometry of these systems crucially depends on the interplay between vdW interactions and wave function hybridisation, making them excellent test cases for vdW-treated DFT. Our results show strong variations in the calculated atomic geometry. While some of the approximations reproduce the experimental structure, this is rather based on \textit{a posteriori} comparison with the ``target results''. General predictive power in vdW-treated DFT is not achieved yet and might require new approaches.